Isomerization and Hydrogenation of cis-2-Butene on Pd …w0.rz-berlin.mpg.de/hjfdb/pdf/515e.pdf · · 2011-11-09Isomerization and Hydrogenation of cis-2-Butene on Pd Model Catalyst

Isomerization and Hydrogenation of cis-2-Butene on Pd Model Catalyst

Bjorn Brandt,† Jan-Henrik Fischer,† Wiebke Ludwig,† Jorg Libuda,‡ Francisco Zaera,§Swetlana Schauermann,*,† and Hans-Joachim Freund†

Fritz-Haber-Institut der Max-Planck-Gesellschaft, Faradayweg 4-6, 14195 Berlin, Germany, Lehrstuhl furPhysikalische Chemie II, Friedrich-Alexander-UniVersitat Erlangen-Nurnberg, Germany, and Department ofChemistry, UniVersity of California, RiVerside, RiVerside, California 92521

ReceiVed: January 9, 2008; ReVised Manuscript ReceiVed: May 7, 2008

The adsorption and kinetics of conversion of cis-2-butene with deuterium on model supported Pd catalyst(Pd/Fe3O4/Pt(111)) were characterized by reflection-absorption infrared spectroscopy (RAIRS), temperature-programmed desorption (TPD), and isothermal molecular beam (MB) experiments. It was found that selectivitytoward cis-trans isomerization and hydrogenation depends critically on the nature of the carbonaceous deposits,which are typically present during reaction on real catalysts. At low temperatures (190-210 K) both reactionpathways were found to proceed on the initially clean surface, but the catalytic activity was observed toquickly vanish, presumably because of the accumulation of hydrocarbon species on the surface. At temperaturesabove 250 K, on the other hand, a sustained catalytic activity toward cis-trans isomerization was observedover long periods of time. Interestingly, no catalytic activity could be sustained for the competing hydrogenationon the initially clean catalyst even at these temperatures. Only when highly dehydrogenated carbonaceousfragments were preadsorbed on the surface was it possible to induce a persistent catalytic activity for thehydrogenation (and also the isomerization) of the alkene on our supported palladium particles. Possible reasonsof this unique vacuum catalytic behavior are discussed, including different spatial requirements for thecompeting reaction pathways and changes in the adsorption state of deuterium on and beneath the surfacemodified by the carbonaceous deposits.

1. Introduction

The promotion of alkene conversions, hydrogenation andisomerization in particular, is required for numerous chemicalprocesses, including fine chemical and pharmaceutical synthesis,petrochemical hydrotreating, and food processing.1,2 Alkenechemistry on metal surfaces has been extensively investigatedin early years using conventional catalytic techniques34,5 andalso more recently by modern surface-science methodologies.67,8

Although the surface-science approach has provided muchinsight into the mechanistic details of the key reaction steps onthe catalyst surface, some important issues remain unresolvedstill. In particular, it is not clear what mechanisms governselectivity in alkene conversions.

When promoted by transition-metal catalysts, alkene conver-sions such as hydrogenation, dehydrogenation, H-D exchange,and isomerization are generally accounted for by the so-calledHoriuti-Polanyi mechanism,9 which proceeds through a seriesof stepwise hydrogenation-dehydrogenation steps. In the par-ticular case of the conversion of 2-butenes with hydrogen ordeuterium, this mechanism involves the formation of a keysurface butyl species via an initial half-hydrogenation of theadsorbed alkene; this species is generally believed to be acommon reaction intermediate for the cis-trans isomerization,H-D exchange, and hydrogenation reaction pathways.2,9–11 Thatbutyl intermediate can then undergo a �-hydride eliminationstep to form an alkene, either the original molecule or one wherecis-trans isomerization or double-bond migration has occurred

(and a hydrogen has been replaced by a deuterium atom ifdeuterium is used as coreactant). Alternatively, the butyl moietycan follow a reductive elimination step with a second coadsorbedhydrogen or deuterium atom to form butane. Dehydrogenationof the adsorbed alkene to other surface species such alkylidynesis also possible.12,13 Figure 1a provides a general schemesummarizing the possible conversions of cis-2-butene withdeuterium.

In most of the studies on reaction selectivities published todate for alkene conversions under UHV conditions, H-Dexchange has been found to dominate over the competinghydrogenation reaction. Thus, extensive H-D exchange has beenobserved for ethene on Ni(100)14 and Rh(111),15 for C2-C4

alkenes on Pd (100),16 for C6 alkenes on Pd(111),17 and forethene,18–20 propene,21 different C4 (1-butene and cis- andtrans-2-butenes),15,22–25 C5 alkenes,26 and cyclic compounds26,27

on Pt(111). In contrast, often only limited if any hydrogenationhas been observed in those systems under the same UHVconditions. Several explanations have been provided for this.Guo and Madix16 ascribed the absence of hydrogenation of thevarious alkenes they studied on Pd(100) to the strong hydrogenadsorption on that surface. Similarly, the weakening ofmetal-hydrogen bonds in the presence of coadsorbates wasinvoked to explain the promotion of alkene hydrogenation onboth a Ni-precovered Pt(111) surface28–30 and on a Fe(100)surface predosed with CO.31,32 Zaera has suggested that the firsthalf-hydrogenation to alkyl species may be the rate-limiting stepin alkene hydrogenation, and that the �-hydride elimination thencompetes favorably against the reductive elimination step withcoadsorbed hydrogen that leads to alkane production.33,34

More recently, the thermal chemistry of alkenes has beenstudied in our group on more realistic model catalysts consisting

* Corresponding author. E-mail: [email protected].† Fritz-Haber-Institut der Max-Planck-Gesellschaft.‡ Friedrich-Alexander-Universitat.§ University of California, Riverside.

J. Phys. Chem. C 2008, 112, 11408–1142011408

10.1021/jp800205j CCC: $40.75 2008 American Chemical SocietyPublished on Web 07/09/2008

of Pd nanoparticles supported on Al2O3/NiAl(110) oxidefilms.35–39 In particular, it has been shown that, at least for etheneand trans-2-pentene, H-D exchange occurs on both Pd(111)single crystal and supported Pd particles, but hydrogenationproceeds only on the nanoparticles. Our experimental resultsstrongly suggest that the formation of weakly bound subsurfacehydrogen species, the accessibility of which is enhanced onnanoscale particles, is the key factor for the hydrogenationreaction to occur.

An additional complication to the understanding of hydro-genation and isomerization processes with alkenes arises fromthe fact that under catalytic conditions those are accompaniedby an early decomposition of some of the reactants and theconcurrent deposition of partly dehydrogenated carbonaceousspecies on the surface.40–42 It has been determined in previousstudies that, indeed, alkene conversions do not take place on

clean metal surfaces but rather on metals covered by thosestrongly bound carbonaceous deposits.43–47 Several stable surfaceintermediates originating from alkene thermal transformationshave also been identified spectroscopically on various singlecrystal surfaces. The most extensive molecule studied to datein this respect is ethylene, from which ethylidene on platinum,48–50

ethylidyne on palladium,51,52 platinum,53–58 rhodium,59,60 ruthe-nium61 and iridium,62 vinyl on nickel,63 palladium,64 andplatinum,65,66 vinylidene on palladium,67,68 and ethynyl onnickel,69–71 platinum,72 palladium73 and rhodium60 have beenidentified. Larger alkylidyne24,74–79 and allylic79–84 species havealso been isolated when starting from heavier hydrocarbons.Those strongly adsorbed carbonaceous species, the alkylidynesin particular, are often considered mere spectators in catalyticreactions, but their exact role in the catalytic mechanism is stillnot known.

Figure 1. (a) Schematic representation of the three competing reaction pathways for cis-2-butene on Pd surfaces being considered in this study,namely: (1) cis-trans isomerization, which is accompanied by H-D exchange in the presence of coadsorbed deuterium, and which produces trans-2-butene-d1 and cis-2-butene-d2 as a mono- and doubly H-D exchanged/isomerized products, correspondingly, (ii) hydrogenation, which leads tothe formation of butane (butane-d2 and butane-d3, the latter as a product of consecutive H-D exchange and deuteration steps), and (iii) dehydrogenation,which results in the formation of hydrocarbon fragments on the surface. (b) Schematic representation of the experimental setup for the molecularbeam experiments. (c) Scanning tunneling microscopy (STM) image (100 nm × 100 nm, inset: 20 nm × 20 nm) of the Pd/Fe3O4/Pt(111) supportedmodel catalyst used in the experiments described here, together with a schematic representation of the structure of the Pd nanoparticles in thatcatalyst.

Isomerization and Hydrogenation of cis-2-Butene J. Phys. Chem. C, Vol. 112, No. 30, 2008 11409

In this contribution, results are presented from our study ofthe kinetics of cis-2-butene conversion with deuterium over amodel catalyst consisting of palladium nanoparticles supportedon a Fe3O4 thin film. Particular emphasis is placed here on theunderstanding of the role of the carbonaceous deposits indefining the selectivity of the catalyst toward different reactionpathways such as hydrogenation and cis-trans isomerization/H-D exchange. Our ability to prepare well-defined supportedmodel catalysts allowed us to design surfaces with a reducedlevel of complexity so they can be probed by surface-sciencetechniques but which still mimic specific features of realcatalysts.85–89 The Fe3O4 thin oxide film was chosen as thesupport here to ensure high stability under reaction conditions.90

A combination of molecular beam methods and infraredreflection-absorption spectroscopy91 has been applied to obtaindetailed isothermal surface kinetic data and to correlate thosewith vibrational spectroscopy evidence for the formation ofdifferent reaction intermediates on both clean Pd particles andsurfaces precovered with specific types of carbonaceous depos-its. The nature of those carbonaceous species is shown tocritically control the activity and selectivity of the cis-2-buteneconversion with deuterium. Particularly, both hydrogenation andisomerization reaction pathways are shown to proceed on theinitially clean metal particles even at low temperatures (190-210K). However, their catalytic activity quickly vanishes underthose circumstances because of the accumulation of surfacehydrocarbon species. At temperatures above 250 K, a sustainedcatalytic activity toward cis-trans isomerization/H-D exchangeis observed, but no steady-state hydrogenation rate could bedetected. Most interestingly, it is shown that under the samereaction conditions a surface precovered with carbon exhibitsa persistent catalytic activity toward both cis-trans isomerizationand hydrogenation. This implies that the carbonaceous speciesresulting from the early decomposition of the reactants are notmerely spectators in alkene conversions, but rather modify theadsorption properties of the surface and critically controlselectivity. Possible reasons of this unique catalytic behavior,including different spatial requirements of the competingreaction pathways and changes in the adsorption state ofdeuterium on the surface modified by the carbonaceous deposits,are discussed. In addition, it is also reported here that alkeneconversion on our model catalytic surfaces can be sustainedover long times under our vacuum conditions, something that,to the best of our knowledge, has not been achieved before.92,93

2. Experimental Section

All molecular beam (MB), temperature programmed desorp-tion (TPD), and reflection-absorption infrared spectroscopy(RAIRS) experiments were performed at the Fritz-Haber-Institut(Berlin) in a UHV apparatus described in detail previously.91

A schematic representation of the setup is shown in Figure 1b.Briefly, this system offers the experimental possibility ofcrossing up to three molecular beams on the sample surface.An effusive, doubly differentially pumped multichannel arraysource was used to supply the D2. This beam was modulatedusing remote-controlled shutters and valves. Beam fluxes forD2 of 3.2 × 1015 molecules cm-2 s-1 were used in theseexperiments. The source was operated at room temperature. Thebeam diameter was chosen such that it exceeded the samplediameter. A supersonic beam, generated by a triply differentiallypumped source and modulated by a solenoid valve and a remote-controlled shutter, was used to dose the cis-2-butene (Aldrich,>99%) at a flux of 5.6 × 1012 molecules cm-2 s-1 (typicalbacking pressure: 1.15 bar). The diameter of this beam was

chosen to be smaller than the sample for the experimentsdiscussed here. Modulation of the beams, in particular that ofthe butene, was performed to differentiate between the productsdesorbing from the sample because of chemical reactions andany other interfering signals such as a slowly growing back-ground observed over the course of the experiments. It shouldalso be noted that the steady-state experiments reported belowwere carried out at temperatures well above those required forthe desorption of the olefins. Therefore, they reflect the kineticsof the chemical reactions on the surface, not that of thedesorption of the hydrocarbons. An automated quadrupole massspectrometer (QMS) system (ABB Extrel) was employed forthe continuous and simultaneous monitoring of the partialpressures of the reactants (2-butene, followed by the signal ofthe C3H5

+ fragment at 41 amu because of experimental reasons)and products (2-butene-d1, C3H4D+ fragment at 42 a.m.u.;2-butene-d2, C3H3D2

+ fragment at 43 a.m.u.; butane-d2,C3H5D2

+ fragment at 45 a.m.u., and butane-d3, C3H4D3+

fragment at 46 a.m.u.). All QMS data have been corrected toaccount for the natural abundance of C13 in all the moleculesfollowed in our experiments. TPD experiments were carried outin the same vacuum system and by using the same QMSinstrumentation. Linear heating ramps for the sample were setto a value of 3.5 K s-1. RAIRS data were acquired by using avacuum Fourier-Transform Infrared (FT-IR) spectrometer (Bruk-er IFS 66v/S) with a spectral resolution of 2 cm-1, using a mid-infrared (MIR) polarizer to select the p-component of the IRlight.

The Pd/Fe3O4 model catalyst was prepared as follows: thethin (∼100 Å) Fe3O4 film was grown on a Pt(111) single crystalsurface by repeated cycles of Fe (>99,99%, Goodfellow)physical vapor deposition and subsequent oxidation (see refs90 and94–96 for details). The cleanliness and quality of theoxide film was checked by RAIRS of adsorbed CO and byLEED. Pd particles (>99.9%, Goodfellow) were grown byphysical vapor deposition using a commercial evaporator (Focus,EFM 3, flux calibrated by a quartz microbalance) while keepingthe sample temperature fixed at 115 K. During Pd evaporation,the sample was biased to +800 V in order to avoid the creationof defects by metal ions. The final Pd coverage used in theseexperiments was 2.7 × 1015 atoms cm-2. The resulting surfaceswere then annealed to 600 K, and stabilized via a few cycles ofoxygen (8 × 10-7 mbar for 1000 s) and CO (8 × 10-7 mbarfor 3000 s) exposures at 500 K before use.90 An STM image ofthe model surface resulting from this preparation is shown inFigure 1c. That surface displays Pd particles with an averagediameter of 7 nm, containing approximately 3000 atoms each,and cover the support uniformly with an island density of about8.3 × 1011 islands cm-2.90 It should be noted that the size ofthe Pd particles appears substantially larger in these STM imagesbecause of the convolution of the signal with that of the tipused as the probe. An accurate estimation of Pd dispersion showsthat only about 20% of the support surface area is covered withnanoparticles.90 The majority of the particles are well-shapedcrystallites grown in the (111) orientation and are predominantlyterminated by (111) facets (∼80%), but a small fraction of (100)facets (∼20%) is also exposed.

3. Results and Discussion

3.1. cis-2-Butene Adsorption and Conversion with Deu-terium on the Fe3O4 Film. As a first step, the adsorptionproperties and the reactivity of the bare Fe3O4 support for cis-2-butene conversion with deuterium was considered. For that,pulsed molecular beam experiments were performed under

11410 J. Phys. Chem. C, Vol. 112, No. 30, 2008 Brandt et al.

isothermal conditions at 200 K. Figure 2a displays a typicalpulse sequence and the time evolution of the partial pressureof the reactant in the vacuum chamber during the run. In thisexperiment, the sample was continuously exposed to a D2 beamduring the complete experiment. Subsequently, the cis-2-butenewas switched on and off, starting after 90 s from the beginningof the D2 dosing. A typical MB sequence included 50 shortbutene pulses (4 s on, 4 s off time) followed by 30 longer pulses(20 s on, 10 s off time). The time evolution of the reactant signaldisplayed in Figure 2a clearly shows two distinct adsorptionmodes for cis-2-butene on the bare Fe3O4 film. First, duringthe first 10 short pulses, the butene sticks effectively to the oxidesurface, leaving behind only a small fraction of the original beamintensity in the gas phase. After that, the gas-phase signal duringexposure to butene increases substantially, indicating a lowersticking probability on the Fe3O4 support. Additional informationon the butene adsorption behavior in these two adsorption modescan be extracted from the analysis of the pulse shapes, shownin detail in Figure 2b. These were calculated by averaging thecis-2-butene traces for the first 10 pulses and the following 40pulses for the “strong” and “weak” adsorption modes, respec-tively. The first adsorption regime is characterized by a squarepulse shape and a high sticking probability, of about 0.65 ascalculated by the ratio of the partial pressure reached here tothe partial pressure at the end of the run, indicating strong andirreversible adsorption of butene species on the surface. Thisstrongly adsorbed state saturates after approximately 2 × 1014

molecules/cm-2 are deposited on the surface. In contrast, thepulse shape of the second adsorption mode deviates clearly fromsuch a square shape. It displays an initial fast increase afterswitching on the butene beam and a slower growth of the signalupon continuing exposure. This behavior suggests that in thismode some butene adsorbs reversibly on the Fe3O4 surface,already covered by the more strongly bonded adsorbate species.It is worth noting that this behavior originates exclusively frombutene adsorption on the Fe3O4 support and does not reflect asaturation process of the chamber walls (this possibility was

ruled out by control experiments where the butene was dosedonto the gold flag instead of the sample). Another importantobservation here is the fact that no gas-phase reaction productswere detected in these isothermal MB experiments, indicatingthat the Fe3O4 surface alone is inactive toward both hydrogena-tion and H-D exchange/isomerization of the alkene (data notshown). The lack of reactivity of the oxide film was corroboratedby TPD experiments carried out on the clean and deuterium-precovered oxide support: only molecular desorption of thereactant was observed in those cases, at 335 K on the cleansurface and between 240 and 290 K on the deuterium-preexposed oxide film (data not shown). This result is in goodagreement with the previously reported lack of C-H bondactivation in alkenes on the Fe3O4 surfaces.97–99

Molecular cis-2-butene adsorption was also confirmed directlyby RAIRS data. The top two traces of Figure 3 show IR spectrarecorded after dosing the alkene on a deuterium-predosed bareoxide surface at 100 K (Figure 3a) and during the isothermalMB experiment at 200 K discussed above (Figure 3b). Fourabsorption bands dominate both spectra: the asymmetric methyldeformation (δas(CH3) at 1450 cm-1), the CdC stretching(ν(CdC) at 1605 cm-1), and the symmetric and asymmetricC-H stretching of the terminal methyl groups (νs(CH3) andνas(CH3) at 2860 and 2915 cm-1, respectively).100–103 Thepresence of the carbon-carbon double bond vibration at 1605cm-1 in particular attests to the previous conclusion concerningthe molecular nature of cis-2-butene adsorption on the oxidesurface. In fact, adsorption of butene may be relatively weak,since the CdC double bond does not undergo the strong sp3

rehybridization due to formation of a di-σ bonded complex

Figure 2. (a) Pulse sequence and time evolution of the cis-2-butenemass spectrometry signal during a typical isothermal pulsed molecularbeam (MB) experiment for the conversion of cis-2-butene with D2 at200 K on our Fe3O4 model support. The sample was first exposed to acontinuous D2 beam (flux 3.2 × 1015 molecules cm-2 s-1), after whicha second modulated cis-2-butene beam (flux 5.6 × 1012 molecules cm-2

s-1) was introduced. Typical MB sequences included 50 short pulses(4 s on, 4 s off time) followed by 30 longer pulses (20 s on, 10 s offtime). (b) Averaged cis-2-butene signal for the first 10 pulses, indicatingstrong irreversible adsorption. (c) Averaged cis-2-butene signal for thefollowing 40 pulses showing a weak reversible adsorption behavior.

Figure 3. RAIR spectra obtained (a) on the Fe3O4 film aftersequentially dosing 280 L of deuterium and 0.85 L of cis-2-butene at100 K, (b) on the Fe3O4 film during the isothermal MB beam experimentat 200 K shown in Figure 2, and (c) on the supported Pd/Fe3O4 catalystafter sequentially dosing 280 L of deuterium and 0.8 L of cis-2-buteneat 100 K.


typically seen for alkenes on metal surfaces.10,13,23,104,105 On theother hand, the vibrational frequency of ν(CdC) shifts by about55 cm-1 compared to its gas-phase value (1660 cm-1), sug-gesting a relative strong interaction of the adsorbed alkene withthe Fe3O4 surface more than what would be expected fromsimple physisorption. Additionally, based on the metal surfaceselection rule (MSSR) that applies to RAIRS with metals,106,107

it may be inferred from the attenuation of the peaks for thesymmetric CH3 stretching (2870 cm-1) and deformation (1375cm-1) as well as the in-plane C-H deformation (1403 cm-1)modes that the adsorption geometry of the alkene is nearly flaton the oxide support.

3.2. Adsorption and Decomposition of cis-2-Butene on PdParticles Supported on the Fe3O4 Film in the Presence ofCoadsorbed Deuterium. Next, the adsorption and decomposi-tion of butene in the presence of deuterium on the Pd particleswas characterized by temperature-programmed desorption (TPD)experiments. Figure 4 displays TPD spectra obtained aftersequentially dosing approximately 280 L of D2 and 0.8 L ofthe alkene at 100 K. Those exposures were chosen to ensurenear saturation of the surface with both atomic deuterium andmolecular butene. Five TPD data sets were obtained sequentiallyby ramping the temperature up to approximately 500 K, coolingdown to 100 K, and redosing the reactants without cleaning orany other treatment of the surface in between experiments inorder to test the effect of the carbonaceous deposits that buildup on the surface on its reactivity.

The H2 TPD trace obtained from the first experiment (Figure4, center panel, top trace) indicates a threshold for hydrocarbondecomposition at about 200 K and a stepwise dehydrogenationwith peaks around 285 and 385 K. It should be kept in mind,however, that the first H2 desorption peak at 200 K might belimited by associative hydrogen desorption, which means thathydrocarbon decomposition may begin even at lower temper-atures. Much more limited decomposition is seen after the firstTPD, but reproducible small peaks are observed in all subse-quent traces with desorption maxima at about 275 and 400 K

(Figure 4, center panel, four bottom traces). The inhibition ofthe dehydrogenation reactions after the initial TPD is ac-companied by an increase in molecular desorption from a weaklybound state. This is manifested by the molecular TPD peaks at145 and 185 K, evolving into a larger feature peaked at 160 Kand a small shoulder around 230 K (Figure 4, left panel).Extensive isomerization and H-D exchange of the alkene onthe surface is also observed: a substantial amount of monodeu-terated trans-2-butene-d1 is produced at 200 K and, to a lesserextent, at 255 K on the clean surface, and detectable C4H7Dproduction is still observed after five TPDs, mostly at 230 K(Figure 4, right). It should be mentioned that doubly exchanged/doubly isomerized cis-2-butene-d2 was also found to desorbaround 200 and 250 K, in peaks that change little with thenumber of TPDs performed, and that deuteration to butane-d2

was observed at 205 and 235 K on the clean surface (data notshown). The hydrogenation (deuteration) reaction is completelysuppressed after the first TPD (data not shown). No otherproducts from butene conversion with deuterium were observed.

The nature of the surface adsorbates originating from cis-2-butene exposure on the supported Pd clusters was investigatedby RAIRS. Figure 3 shows a comparison of the RAIRS dataobtained after cis-2-butene exposure on the pure Fe3O4 film(Figure 3a) versus on the Pd particles supported on this oxidesurface (Figure 3c). Both surfaces were pretreated with 280 Lof deuterium before exposure to the alkene, and the spectra weretaken at 100 K. The IR spectra obtained on the supported metalparticles show pronounced differences as compared to thepristine oxide film: whereas two of the main features, theasymmetric methyl deformation δas(CH3) at 1450 cm-1 andasymmetric C-H stretching of the terminal methyl groupsνas(CH3) at 2900 cm-1, remain nearly the same (the νas(CH3)peak shifts slightly from the original 2915 cm-1 value), the thirdprominent vibration for the CdC stretching, ν(CdC), at 1605cm-1, vanishes completely in the spectra on Pd particles. Thisindicates a strong rehybridization of the double bond and theformation of a di-σ complex on the metal surface. It is worth

Figure 4. cis-2-Butene (left panel), hydrogen (center panel), and trans-2-butene-d1 (right panel) temperature programmed desorption (TPD) tracesobtained after sequential dosing of 280 L of D2 and 0.8 L of cis-2-butene at 100 K on our Pd/Fe3O4 model catalyst. Data are reported for fiveconsecutive experiments carried out by ramping the temperature to up to 500 K each time and dosing the reactants on the surface resulting fromthe previous experiment without any intermediate treatment. Particularly noteworthy here is the persistent signal obtained for trans-2-butene-d1

production (right panel) after the repeated experiments, indicating sustained reactivity toward isomerization even after the build up of a carbonaceouslayer on the surface.


noting that even though, according to STM data,90 about 80%of the Fe3O4 surface is still exposed in the system with thesupported Pd particles, no support-related butene species wereobserved in the IR spectra. This fact suggests that either theadsorption properties of the support are strongly modified afterPd deposition and stabilization under oxidizing conditions, orthat butene diffuses rapidly from the support to the metalparticles, where it binds more strongly.

In light of these results it is concluded that decomposition ofcis-2-butene on the supported Pd particles proceeds in a similarmanner as on Pt and Pd single-crystal surfaces.16,22,23,25,75,76,108–110

In particular, there is good agreement between the hydrogenTPD traces obtained here and those reported in the literaturefrom single crystals in terms of both peak positions and relativeintensities.22,75 Additionally, the low-temperature RAIRS dataobtained on the Pd particles in this study are consistent withthe formation of a di-σ surface species in the submonolayerregime, in the same way as previously observed spectroscopi-cally on Pt23,75,103 and Ru.110 Based on these data as well as onprevious reports on similar systems10,104,108 we may speculatethat cis-2-butene adsorbs on the Pd particles via the formationof a di-σ bound species in the submonolayer coverage around100 K, and dehydrogenates to a di-σ bonded 2-butyne speciesat room temperature and to highly dehydrogenated carbonaceousdeposits (presumably C4H2) above 400 K.

3.3. H-D Exchange, Isomerization, and Hydrogenation ofcis-2-Butene on Pd Particles Supported on the Fe3O4 Film.As mentioned above in reference to the TPD results in Figure4c, the thermal activation of cis-2-butene coadsorbed withdeuterium on our Pd model catalyst leads to extensive isomer-ization and H-D exchange of the alkene. In addition to thesubstantial production of monodeuterated trans-2-butene at 200and 255 K on the clean surface and at 230 K after the firstTPD, some doubly exchanged/doubly isomerized cis-2-buteneis also made at approximately the same temperatures, anddeuteration to butane-d2 is observed at 205 and 235 K on theclean surface. These results are also, in general terms, inagreement with previous work on Pd(100)16 and Pt(111)22,25

single crystals. While no hydrogenation products have beendetected for any of those alkenes on either surface, extensiveH-D exchange (accompanied by cis-trans isomerization in thecase of the 2-butenes) has been reported both for 1- and2-butenes. Also, cis- and trans-2-butenes were shown to formonly mono and doubly deuterium-substituted species on Pt(111),presumably because only the inner hydrogen atoms can beexchanged; this observation is in full agreement with ourexperiments on the supported Pd particles. However, there arealso some important differences in the alkene conversion onthe metal clusters versus the single crystals. For example, themajority of the H-D exchange/isomerization products in the firstTPD run in this study desorbs at temperatures significantly lowerthan those reported for Pt(111) (200 vs 260 K). Also, eventhough the low-temperature H-D exchange channel seen inFigure 4 shuts off on the surface pretreated with hydrocarbons(that is, after the first TPD run), persistent H-D exchange isstill observed at 230 K in all of the following TPD experiments.This shows that the surface is capable of maintaining its H-Dexchange activity in the presence of large amounts of highlydehydrogenated carbonaceous deposits, something that is alsonot typical on single crystals.

The observation of persistent catalytic activity on the sup-ported Pd particles suggests that free metal sites are still exposedon the surface precovered with highly dehydrogenated carbon-aceous deposits. To verify this hypothesis, the availability of

metal surface sites for reaction was evaluated by performingCO adsorption experiments. The RAIRS spectra obtained onthe clean and C-precontaminated Pd particles are shown inFigure 5. The CO spectrum for the clean surface (bottom trace)is dominated by an intense and sharp peak centered around 1978cm-1, associated with Pd sites on the particle edges and cornersand to a lesser extent with (100) facets.111,112 The broad featuresbetween 1960 and 1800 cm-1 are attributed to CO adsorptionon the regular (111) facets representing the majority of thesurface sites.111,112 It should be pointed out that the intensitiesof the IR signals are expected to be strongly modified by dipole-coupling effects113 and, as a consequence, to not directly reflectthe relative abundance of the corresponding sites. In theparticular case of the Pd particles, the feature at high frequencycorresponding to the low-coordinated sites (edges, corners, etc.)is likely to gain intensity at the expense of the absorption signalat lower frequency related to the regular (111) facets. In anycase, the main effect of the deposition of carbonaceous speciesduring the TPD runs reported here is the complete disappearanceof the 1978 cm-1 peak; the spectral features in the low-frequencyregion remain nearly unchanged. Additionally, a new peakgrows at 2095 cm-1 associated with on-top adsorption sites.These changes indicate preferential adsorption of carbonaceousdeposits on the edges, corners, and (100) facets, and the factthat regular terraces remain considerably less contaminated. Asimilar behavior has previously been observed for carbondeposits generated by methanol decomposition.114,115 RecentDFT calculations also support preferential adsorption of carbonat the edges of Pd clusters and suggest that part of the carbonatoms may be located in subsurface sites.116

Another interesting observation here is the ability of theinitially clean supported particles to hydrogenate the double bondof the olefin, a reaction that is not seen on palladium singlecrystals. It should be noted that while alkenes tend to desorb

Figure 5. CO RAIRS titration spectra for Pd/Fe3O4/Pt(111) surfacesclean (bottom) and after the first TPD experiment shown in Figure 4,which leads to the deposition of highly dehydrogenated carbonaceousspecies. The surface was saturated with CO at room temperature andcooled to 100 K before data acquisition. The data indicate that carbondeposition on the palladium particles leads mainly to blocking of bothlow-coordination defect (edges, corners) sites and bridge sites on (100)facets. The majority of the surface sites on the regular (111) terracesremain unaffected.


molecularly rather than hydrogenate on single crystals,38,117

extensive conversion has been reported already for ethylene andtrans-2-pentene hydrogenation on Pd particles dispersed on thinoxide films.35,39 This difference has been ascribed to the abilityof the dispersed particles to absorb highly reactive hydrogenatoms in a weakly bound subsurface state.37

The significant desorption of H-D exchange products in theTPD experiments described above after deposition of hydro-carbon fragments on the surface strongly suggests that it maybe possible to sustain steady-state conversion on these surfaces.This hypothesis was indeed confirmed by the molecular beamexperiments described next. Additionally, the differences in thecatalytic behavior between the initially clean Pd particles andthe particles precovered with hydrocarbon fragments indicatethat not only activity but also selectivity depends strongly onthe nature of the carbonaceous species present on the surfaceduring the catalytic reaction.

3.4. cis-2-butene conversion with deuterium on Pd/Fe3O4:isothermal molecular beam experiments. In order to furtherinvestigate the activity and selectivity of the Pd/Fe3O4 modelcatalyst for cis-2-butene conversion with deuterium, and toexplore the effect of the carbonaceous deposits, pulsed isother-mal molecular beam experiments were performed by applyingthe same D2 and cis-2-butene dosing scheme used for the studieson the bare Fe3O4 support (see Section 3.2). Figure 6 displaysthe reaction rates of both reaction pathways, H-D exchange/isomerization and hydrogenation, obtained at 200 K on theinitially clean surface (lower 4 traces), together with the tracefor the reactant uptake during the MB run (uppermost trace).In general, the cis-2-butene uptake here exhibits a similarbehavior to that observed on the bare Fe3O4 support, with thetwo distinct adsorption modes identified above: (i) a strongirreversible adsorption at the beginning of the exposure, with ahigh sticking probability (S ) 0.65) and a square pulse shape,and (ii) a subsequent weaker and reversible adsorption, withthe surface concentration of the alkene increasing upon admis-

sion of the beam and decreasing after beam termination.However, initial saturation of the supported catalyst with thestrongly bound hydrocarbon species is reached almost twice asfast as in the case of the bare Fe3O4 film. A number of reasonsmay account for this fact. First, the adsorption properties ofthe support might be modified as a result of Pd deposition andstabilization under oxidizing conditions in such a way thatconsiderably less butene can be adsorbed on the support itselfas compared to the pristine oxide film. However, this effect alonecannot solely account for the length of the induction period onthe supported catalyst. In fact, the length of the induction perioddecreases further with increasing temperature (see Figure 7).This suggests that the build-up of the threshold coverage ofhydrocarbon species during the induction period may involvethermally activated steps taking place on the Pd particles suchas partial dehydrogenation of the butene. Apparently, such partlydehydrogenated hydrocarbon species may be easier formed atelevated temperatures, and hence rapidly saturate the surface.It should be pointed out, however, that the hydrocarbon uptakeobserved in this induction period may also include someadsorption of molecular butene, in particular at the later stagesof the dose, after the formation of partly dehydrogenatedhydrocarbon residues. The relative fractions of carbonaceousdeposits versus molecularly adsorbed olefin resulting from theuptake during the induction period are determined by the relativemagnitudes of the rates of dehydrogenation versus adsorption-desorption equilibrium of molecular butene at a given temper-ature. Therefore the length of the induction period and therelative abundance of the different hydrocarbon species formedon the surface are expected to depend in complex way on thereaction temperature, a way not easy to unravel from ourexperimental data.

Also, in contrast to what was observed on the pristine oxidesurface, significant conversion of the cis-2-butene is seen onthe Pd/Fe3O4 catalyst after the initial adsorption of the butenedescribed above. The formation rates of trans-2-butene-d1, cis-

Figure 6. Results from an isothermal kinetic MB experiment carriedout at 200 K on Pd/Fe3O4 by using the same pulse scheme describedin Figure 2. Shown is the time evolution of both the reactant (cis-2-butene) and the products (trans-2-butene-d1, cis-2-butene-d2, butane-d2, and butane-d3). The reaction rates display a pronounced inductionperiod followed by a transient regime of high activity toward both H-Dexchange/isomerization and hydrogenation reaction pathways. However,after a few hundreds of seconds, all reaction rates return to zero.

Figure 7. Results from MB experiments carried out on Pd/Fe3O4 atdifferent temperatures in the range from 190 to 260 K, again using theMB pulse scheme described in Figure 2. Shown is the time evolutionof the product of cis-trans isomerization (trans-2-butene-d1) for eachtemperature, together with that of the reactant (cis-2-butene, bottomtrace). Below 210 K the reaction rates are high at the beginning butreturn to zero after prolonged exposures. Above 250 K, on the otherhand, there is a shorter induction period, and sustained catalytic activitytoward isomerization is possible.


2-butene-d2, and butane-d2 and -d3 at 200 K all exhibit aninduction period of about 40 s (corresponding to 20 s of totalexposure time in the 5 initial pulses), coinciding with theduration of the strong adsorption regime seen for the reactant(Figure 6). This fact indicates that the initial strong adsorptionmode on the supported system involves the buildup of athreshold coverage of butene and/or other hydrocarbon specieson the Pd particles that do not directly participate in but preparethe surface for the subsequent reactions. Only after the metalsurface is saturated with those spectator hydrocarbon speciesthe butene catalytic conversion becomes possible.

Once the conversion starts, the rates of all the reactionpathways listed above show a distinct transient behavior, withcomparable rates that reach maxima at about 80 s and decreaseto zero after about 300-400 s for H-D exchange/isomerizationand after 200 s for deuteration.

Figure 7 displays the reaction rates of trans-2-butene-d1

production for a series of temperatures between 190 and 260K. Here, it should be emphasized that the reported experimentswere carried out at temperatures well above those required forthe desorption of the olefins. Therefore, they reflect the kineticsof the chemical reactions on the surface, not that of thedesorption of the hydrocarbons. Similar kinetics are observedfor the production of trans-2-butene-d1 between 190 and 210K, with the transient behavior comprising an induction period,an initial high activity, and an asymptotic decay in reaction rateto zero after a few hundreds of seconds. However, the reactivitypattern markedly changes when the reaction is carried out at250 and 260 K. First, the surface follows a considerably shorterinduction period than in the low-temperature range, of only 8 sat 250 K (compared to 20 s at 200 K). In addition, there is apersistent catalytic activity toward H-D exchange/isomerizationat 250-260 K even after prolonged periods of time. In fact,the activity at these temperatures was probed for much longertime scales than that displayed in Figure 7, and several turnoverswere obtained (over a period of more than 2 h) without anysignificant reduction in the steady-state reaction rates ofcis-trans isomerization. This is perhaps the most importantobservation from this experiment, because, to the best of ourknowledge, sustained catalytic activity has not been achievedunder vacuum before, not even in molecular beam experimentswith large excesses of hydrogen in the gas mixture.92,93

The substantial decrease in the induction period seen at250-260 K suggests the formation of a specific type of stronglyadsorbed hydrocarbon species on the Pd particles at thesetemperatures, which involves a thermally activated step. It isimportant to note that this transition roughly coincide with theonset of hydrogen desorption seen in TPD, which is associatedwith the first step of cis-2-butene decomposition on the Pdsurface. Even though it is not clear from the TPD data if thehydrogen released at 250 K is kinetically limited by alkenedissociation or by recombinative hydrogen desorption, the abruptchange in the length of the induction period in the MB runsstrongly suggests that butene decomposition takes place in thistemperature range. Based on information from the literature, itcan be speculated that the first dissociation step presumablyinvolves abstraction of the inner hydrogen atoms in the adsorbed2-butene and leads to the formation of di-σ bonded 2-butynespecies.103

As the possible reasons for the transition from limitedconversion to the persistent catalytic activity seen around 250K in Figure 7 are discussed, it is worth remembering that thelack of sustained conversion in vacuum experiments is usuallyascribed to the strong inhibition of dissociative hydrogen/

deuterium adsorption on the metal surfaces induced by thepresence of strongly bonded hydrocarbon species.2,44 Consistentwith this is the fact that both hydrogenation and isomerizationrates for cis-2-butene on Pt(111) follow a near-first-orderdependence on the hydrogen pressure and a zero dependenceon the hydrocarbon pressure.118 On the basis of these consid-erations, it appears that in the low-temperature region thereactions observed in the initial stages of the MB runs involveonly the hydrogen (or deuterium) adsorbed at the very beginningof those runs, before the surface is exposed to the butene beam.Once the hydrocarbon exposure starts, however, the buteneappears to form strongly adsorbed hydrocarbon surface species(most likely a di-σ species) without any deuterium consumption,until a threshold coverage of those species is reached. Afterthis induction period, additional butene may react with thedeuterium already present on the surface, hence the high initialreaction rates for both H-D exchange/isomerization and hydro-genation reaction pathways seen in Figure 7. However, furtherdissociative adsorption of deuterium is prevented by thehydrocarbon deposits formed during early butene decomposition,so the activity of the surface vanishes after the initially availabledeuterium is consumed.

Following the same reasoning, it can be said that the sustainedcatalytic activity toward isomerization/H-D exchange observedabove 250 K attests to the ability of the Pd particles todissociatively adsorb deuterium at these temperatures, incompetition with butene adsorption and in the presence of partlydissociated hydrocarbon species on the surface. This was in factproven directly in isotope switching experiments, where thehydrogenation and H-D exchange reactions could be reversiblyturned on and off by either interchanging H2 and D2 beams orswitching either one off and back on (data not shown). Twopossible reasons might account for this phenomenon. First, themore strongly dehydrogenated hydrocarbon spectator speciesthat form on the surface above 250 K might have a smallerfootprint, and may leave some open metal sites available fordeuterium dissociative adsorption. Second, faster butene de-sorption (as compared to the low temperature case) may resultin a lower steady-state coverage of the reversibly adsorbedbutene, diminishing the potential poisoning effect of thosehydrocarbon species toward the dissociative adsorption ofdeuterium. Both tendencies result in the reduced poisoning ofthe surface by hydrocarbons, and to the sustained ability of Pdparticles to dissociatively adsorbed deuterium under reactionconditions. These results demonstrate for the first time, to thebest of our knowledge, that alkene conversion can be maintainedover long times under the vacuum conditions92,93 and arguestrongly for the power of using realistic models such as ours tosimulate real catalysis; it appears that similar hydrogen uptakeis not possible on single-crystal surfaces covered with carbon-aceous deposits.79 It is worth noting, though, that the reactionprobabilities measured under our conditions are on the order ofa few percent. For comparison, a typical alkene catalytichydrogenation process carried out around room temperature andunder atmospheric pressures exhibits reaction probabilities onthe order of 10-5-10-10.2,5,105 At present we do not have a fullysatisfactory explanation for the unique high activity of our modelcatalyst.

3.5. Selectivity between Isomerization and Hydrogenationon Pd/Fe3O4. The TPD data discussed in section 3.3 indicatethat not only activity but also selectivity toward different reactionpathways might be affected by the presence of differentcarbonaceous species on the surface. To study this dependencyin more detail, comparative pulsed isothermal MB experiments


were performed at 260 K on both initially clean and carbon-precovered Pd particles. The Pd particles precovered with carbonwere prepared following the procedure used in the TPDexperiments, namely, by dosing 280 L of D2 and 0.85 L of thealkene sequentially at 100 K and then annealing to ap-proximately 500 K. As discussed earlier, this treatment produceshighly dehydrogenated carbonaceous species containing only afew if any hydrogen atoms. For the sake of simplicity, we willrefer to these species as carbon in the following discussion.

Figure 8 shows the time evolution of the signals for theoriginal cis-2-butene and for the four main reaction products(trans-2-butene-d1, cis-2-butene-d2, butane-d2, and butane-d3)as a function of reaction time in experiments carried out at 260K starting with the clean (left) and carbon-covered (right)surfaces. As discussed in the previous section, on the initiallyclean surface, a transient behavior is seen for all products,consisting of an induction period in which no reaction productsare detected (and only the uptake of the cis-2-butene is observed)followed by a period of high activity for both isomerization/H-D exchange and deuteration. However, whereas the isomer-ization rates to trans-2-butene-d1 and cis-2-butene-d2 aresustained over longer periods of time at this temperature, thedeuteration rates to butane-d2 and butane-d3 return to zero.

Interestingly, the surface precovered with carbon follows aninduction period for the uptake of butene similar to that seenwith the clean surface. This indicates that carbon contaminationdoes not significantly change the total surface area accessiblefor butene adsorption, an observation corroborated by the CO-titration IR spectra obtained on both clean and carbon-precovered Pd particles (see Figure 5): carbon contaminationwas found to affect mostly the defect sites (edges and cornersof the Pd particles) without reducing adsorption capacity of the(111) facets, which represent the majority of the surface sites.In addition, recent theoretical calculations on Pd clusters showthat the binding energy of carbon on edges is about the same

as in the subsurface region near those edges.116 It is thereforeeven possible that the carbon in the 500 K-pretreated surfacesdiffuses to the inside of the particle through its edges.

Also surprising here is the observation that not only isomer-ization but also hydrogenation to both butane-d2 and butane-d3

can be sustained under steady-state conditions on the carbon-covered surface. Additionally, the two competing reactionpathways were found to exhibit very different transient behavior.Figure 9 shows the time evolution of the reaction rates forisomerization and hydrogenation, together with the time evolu-tion of the cis-2-butene in the chamber, during the individualcis-2-butene long pulses on the initially clean (left) and carbon-precovered (right) surfaces (averaged over the last 30 pulses).The transient kinetic data show that both isomerization ratessimply follow the time evolution of the reactant, suggesting thatthey are limited by the rate of cis-2-butene adsorption underthe given conditions. In contrast, both hydrogenation rates onthe C-precovered surface are high at the beginning of the cis-2-butene exposure but decrease considerably over the durationof the pulse. This behavior can be explained by noting that thesurface is saturated with deuterium atoms before the start ofthe alkene exposure (because the deuterium beam is kept oncontinuously) and hence is highly reactive toward deuterationas soon as the alkene becomes available at the beginning of thebutene pulse. However, as the reaction proceeds, that deuteriumis consumed, and the initially high surface coverage of deuteriumdecreases to a new lower steady-state level because of the partialpoisoning of the surface by butene.

An additional consideration here is the fact that, accordingto the Horiuti-Polanyi mechanism, both H-D exchange/isomerization and hydrogenation (deuteration) reactions involvetwo elementary surface steps, the first of which is the formationof a common surface 2-butyl intermediate.9,10 The fast isomer-ization and deuteration seen at the beginning of each pulsesuggest that butyl formation is fast, and that the selectivity is

Figure 8. Results from isothermal pulsed molecular beam (MB) experiments on the conversion of cis-2-butene with D2 at 260 K on initially clean(a) and C-precovered (b) Pd/Fe3O4 model catalysts. The latter sample was prepared by performing one TPD cycle as shown in Figure 4 prior to theMB run. Data are reported here for the evolution of the reaction rates as a function of time for the reactant (cis-2-butene) and for the main fourproducts (trans-2-butene-d1, cis-2-butene-d2, butane-d2, and butane-d3). They indicate that while on the initially clean surface only sustainedisomerization is possible (left panel), both isomerization and hydrogenation production persist on the surface precovered with carbon deposits (rightpanel).


defined by competition between the two subsequent reactions,�-hydride elimination for the exchange versus reductive elimi-nation with a coadsorbed deuterium atom for the deuteration.The former step does not require any additional reactants, andtherefore follows closely the kinetics of butyl formation, whichin turn depends on the uptake of the butene. The latter, on theother hand, requires an additional deuterium surface atom. Theobserved transient behavior suggests that the hydrogenation rateis controlled by the number of accessible deuterium atoms, thatis, by the dissociative adsorption of deuterium.

It is particularly puzzling to see such dramatic change inselectivity with changes in the nature of the carbonaceousdeposits present on the surface. Here it should be kept in mindthat even when starting with a clean surface, strongly bondedcarbonaceous deposits do form and persist after the initialinduction period in the MB isothermal experiments. So it is thedifferent nature of those deposits when annealing to ∼500 Krather than their mere presence on the surface what leads to thepromotion of the hydrogenation reactions. In fact, annealing thepredeposited adsorbed butene to only ∼350 K is not sufficientto achieve this behavior (data not shown). Based on the TPDdata shown in Figure 4, we suggest that the more active surfacemay contain highly dehydrogenated surface species, perhapseven carbon alone in carbidic form, crucially, that carbon maybe present in the subsurface (see below).

The predeposited carbon must be capable of performing atleast two functions in order to help preserve the catalytic activityof the surface. On the one hand, it must block extensive alkenedehydrogenation pathways, at least when a set coverage ofcarbonaceous deposits on the surface is reached, otherwise thesurface would eventually be poisoned by those. On the other,it must also leave some exposed metal sites available fordeuterium adsorption and alkene conversion. Both are possibleif the different reactions require catalytic sites with differentPd ensemble sizes, or if they occur on different surface sites.That way, the carbon can be seen as selectively promoting some

reactions at the expense of others, at least in relative terms.Indeed, dehydrogenation steps are believed to require largenumbers of metal atoms, so those could be poisoned easily.2

Hydrogenation, either to alkyl species or to alkanes, may bepossible on smaller sites and, therefore, still be feasible onpartially carbon-covered surfaces. Even smaller sites may becapable of sustaining the H-D exchange path; hence, thesustained activity seen for that reaction even when starting fromthe clean surface (where steady-state alkane production is notpossible). Certainly, different adsorption requirements areapplicable to alkenes compared to those for hydrogen/deuterium,as manifested by the different results that are often obtained incoadsorption TPD experiments depending on the order in whichthe two reactants are dosed on the surface.7,14 Moreover,adsorption of different hydrocarbon fragments requires differentadsorption geometries, and hydrogenation and dehydrogenationsteps also involve different ensembles of surface atoms.8,10,119

Most of the results reported to date on surface-science studieson model systems involving hydrocarbon chemistry are sup-portive of the idea of selectivity in hydrogenation versusdehydrogenation steps being dependent on the size and natureof the surface sites available for reaction.

Alternatively, it could be argued that in the case of thepretreatment of the Pd particles at 500 K the nature of the surfacechanges because of the presence of carbon in the subsurface.As mentioned above, there are at least three reasons to believethat the surface in that case is fairly clean, and that most of thecarbon initially deposited by butene adsorption and decomposi-tion diffuses into the subsurface region after annealing to 500K: (i) the CO RAIRS titration spectra in Figure 5 show thatonly defect sites are affected by dosing butene on the surfaceand annealing to 500 K, the majority of the surface sites on the(111) facets appear to remain free; (ii) the induction period forthe uptake of the olefin in the MB experiments is the same forboth clean and pretreated surfaces (Figure 8); and (iii) the steady-state isomerization rates for the mono H-D exchanged productare similar on the initially clean and C-precovered surfaces.Perhaps the subsurface carbon modifies the electronic natureof the surface palladium atoms, and that affects the behavior ofthe catalyst toward alkene hydrogenation and/or dehydrogena-tion (but apparently not toward H-D exchange or double bondisomerization) reactions. It should be mentioned that Pd surfacespretreated by adsorbing butene at low temperature and annealingto 350 instead of 500 K do not facilitate the catalytichydrogenation of the olefin, and behave quite similarly to thecase of starting with the clean surface. Clearly, in that case thecarbonaceous deposits formed by the pretreatment are quitesimilar to those that form during the induction period ofthe MB experiments and do not modify the performance of thesurface in the way that the ”carbidic” subsurface species do.

A third possibility to account for the changes in selectivityreported here with surface pretreatment is that the hydrogenationcatalytic activity induced by strongly bound carbonaceousspecies on Pd particles is due to a change of the hydrogenadsorption state in the presence of the deposited carbon.Previously, the origin of the hydrogenation activity observedon dispersed Pd particles but not on single crystals has beenascribed to the unique ability of the small metal clusters to bindhydrogen in a special weakly bound state.37 Recent NRA(nuclear reaction analysis) experiments on the hydrogen distri-bution in the dispersed Pd particles supported on an oxide filmprovided clear experimental proof for those weakly boundhydrogen atoms being a subsurface/bulk species.120 Our ex-perimental data support this idea that two different forms of

Figure 9. Average reaction rates as a function of time during eachcycle in the latter part of the isothermal MB experiments reported inFigure 8. Data are again contrasted for the initially clean (left panel)versus carbon-precovered (right panel) Pd particles. The rates werecalculated by averaging the steady-state reaction rates from the last 30long pulses of the experiment.


hydrogen (deuterium) may be involved in the competingisomerization and hydrogenation reaction pathways. This isillustrated by the results in Figure 8 in particular those for thereaction rates at 260 K for isomerization to trans-2-butene-d1

and deuteration to butane-d2 on the initially clean particles. Thedata show that when the reaction starts, after the clean particleshave been saturated with deuterium at the beginning of theexperiment, both reaction pathways display about the samereaction rates. However, the hydrogenation rate drops soonafterward, and reaches a value of zero under steady-stateconditions. The isomerization rate, on the other hand, remainsas high as at the beginning of the experiment. The two reactionsshare a common hydrocarbon reaction intermediate, a butylsurface species produced by the half-hydrogenation of theadsorbed olefin (see the reaction scheme in Figure 1), which isformed readily on the surface as indicated by the sustainabilityof the H-D exchange reaction. Therefore all differences in thereaction kinetics of the H-D exchange/isomerization and hy-drogenation pathways are most likely related to their differentrequirements in deuterium atoms. The observed selectivesuppression of only the hydrogenation pathway on the initiallyclean surface under steady-state conditions is not likely to beexplained simply by changes in hydrogen (deuterium) surfacecoverages since both reactions require deuterium additions andexhibit similar reaction orders on deuterium (hydrogen) pres-sures.118 More likely, the vanishing hydrogenation activity canbe rationalized by assuming that a special type of deuteriumatoms is required for the second hydrogenation step (conversionof butyl to butane). That special deuterium species may formduring the initial deuterium uptake, but become depleted duringthe course of the reaction. In line with the previous observationswe assume that this special type of deuterium atoms might bea weakly bound subsurface/bulk species. It could be argued thatdeposition of strongly bound carbonaceous species weakens themetal-deuterium bond and makes it possible to replenish thepopulation of such special deuterium under steady-state condi-tions, resulting in the sustained hydrogenation conversion.

Indeed, TPD measurements carried out on the clean andcarbon precovered Pd particles show that the distribution ofhydrogen atoms is strongly shifted toward lower temperatureson surfaces precovered with carbon, suggesting that such weaklybonded species may be the ones responsible for the productionof the alkane.36 Currently, the experiments are under way, wherethe distribution of the hydrogen species under isothermalreaction conditions is investigated by means of NRA both onthe clean and carbon-precovered Pd particles, which will be asubject of a publication in the near future.121

Finally, it should be pointed out that the proposed explana-tions of the observed unique catalytic behavior seen here withthe supported palladium model system are speculative, andshould be considered rather as working hypotheses to be testedin further experiments.

4. Conclusions

The catalytic conversion of cis-2-butene on a well-definedmodel supported catalyst was investigated under UHV condi-tions by using a combination of RAIRS, TPD, and molecularbeam experiments. Emphasis was placed on exploring the roleof the carbonaceous deposits typically present under catalyticreaction conditions on the activity and selectivity towarddifferent reaction pathways, in particular cis-trans isomerizationversus hydrogenation.

The Pd model catalysts used consisted of Pd particles of 7nm average size with predominant (111) facets (but also

exposing a small fraction of (100) surfaces) deposited on a well-ordered Fe3O4 film grown on Pt(111) as a support. The followingkey conclusions were reached:

1. cis-2-butene adsorbs molecularly on the pristine Fe3O4 filmeither clean or after predosing deuterium. Two adsorption modeswere identified in molecular beam experiments, a strong andirreversible state that populates first, at the start of the olefinuptake, and a second, weaker and reversible adsorption regimethat develops afterward. No catalytic activity toward theconversion of cis-2-butene with deuterium was observed on theFe3O4 support.

2. Starting with a Pd/Fe3O4 catalyst predosed with deuterium,thermal activation of adsorbed cis-2-butene in TPD experimentsleads to several reaction pathways, including dehydrogenation,cis-trans isomerization (which is accompanied by H-D ex-change), and hydrogenation. Dehydrogenation of the alkene wasalso found to proceed, in several consecutive steps. Carbondeposition on the Pd particles by annealing of the surfacecovered with the olefin at 500 K leads to inhibition of thedehydrogenation reaction pathways, and increases moleculardesorption of the reactant.

3. The isothermal uptake of cis-2-butene on Pd/Fe3O4 goesthrough an induction period followed by a period of highreactivity toward both cis-trans isomerization and hydrogena-tion. The induction period is associated with accumulation ofstrongly adsorbed hydrocarbon species, which might includemolecularly adsorbed butene and/or partly dehydrogenatedbutene derivatives depending on the reaction temperature.

4. At low (190-210 K) temperatures, the initial reactivitytransient described above is followed by a decrease in allreaction rates until returning to zero after a few hundreds ofseconds. This behavior is likely to be the consequence of thestrong inhibition of the dissociative deuterium chemisorptionby the presence build-up of bulky hydrocarbon species, mostlikely di-σ bonded butene.

5. At temperatures above 250 K, sustained catalytic activitytoward cis-trans isomerization is possible. This is, to the bestof our knowledge, the first time that such sustained catalyticbehavior has been seen under vacuum conditions. The onset ofthe steady-state reactivity is accompanied by the reduction ofthe induction period in the initial olefin uptake. This behaviorsuggests that (i) the initial accumulation of the strongly boundhydrocarbon species that form on the surface during theinduction period above 250 K is a thermally activated process,which produces partly dehydrogenated carbonaceous deposits,probably exhibiting a smaller surface area footprint, and (ii)the catalyst surface is still able to dissociatively adsorb deuteriumabove 250 K, even in competition with the molecular buteneadsorption and in the presence of surface butene derivatives.However, despite of the observed sustained catalytic activitytoward cis-trans isomerization, the hydrogenation reactionpathway cannot be maintained on the initially clean surface.

6. On surfaces pretreated by adsorbing cis-2-butene at 100K and annealing at 500 K, a different catalytic behavior is seenwhere both reaction pathways, cis-trans isomerization andhydrogenation, are sustained over long periods of time. Clearly,carbon deposition aids in the sustainability of the hydrogenationreaction pathways, and thus critically controls activity andselectivity. Several pieces of evidence point to the wideavailability of surface sites, in particular in the (111) facets,and to the presence of subsurface carbon in this case.

A discussion of the possible reasons for the observedphenomena was provided. The facilitation of the catalytichydrogenation reaction pathway in particular could be possibly


ascribed to a combination of the following factors: (i) differentspace requirements for the H-D exchange/isomerization andhydrogenation competing reaction pathways, (ii) an electronicmodification of the surface by the presence of subsurface carbon,and/or (iii) changes in the distribution of deuterium betweenweakly and strongly bound adsorption states, which may exhibitdifferent reactivity toward hydrogenation.

Acknowledgment. Financial support by the following agen-cies is gratefully acknowledged: Deutsche Forschungsgemein-schaft, Fonds der Chemischen Industrie, and the U.S. NationalScience Foundation. F.Z. thanks the Alexander-von-HumboldFoundation for a senior scientist award. Finally, we thank Prof.R.J. Madix for helpful discussions.

References and Notes

(1) Ponec, V.; Bond, G. C. Catalysis by Metals and Alloys; Elsevier:Amsterdam, 1995.

(2) Bond, G. C. Metal-Catalysed Reactions of Hydrocarbons; SpringerScience: New York, 2005.

(3) Taylor, T. I. In Catalysis; Emmett, P. H., Ed.; Reinhold: NewYork, 1957: Vol. 5, pp 257-403.

(4) Kemball, C. In AdVances in Catalysis and Related Subjects; Eley,D. D., Selwood, P. W., Weisz, P. B., Eds.; Academic Press: New York,1959; Vol. 11, pp 223-262.

(5) Horiuti, J.; Miyahara, K. Hydrogenation of Ethylene on MetallicCatalysts. Report NSRDS-NBC No. 13; National Bureau of Standards:Gaithersberg, MD, 1968.

(6) Somorjai, G. A. Introduction to Surface Science Chemistry andCatalysis; John Wiley & Sons: New York, 1994.

(7) Zaera, F. Prog. Surf. Sci. 2001, 69, 1.(8) Ma, Z.; Zaera, F. Surf. Sci. Rep. 2006, 61, 229.(9) Horiuti, J.; Polanyi, M. Trans. Faraday. Soc. 1934, 30, 1164.

(10) Zaera, F. Chem. ReV. 1995, 95, 2651.(11) Zaera, F. Acc. Chem. Res. 1992, 25, 260.(12) Koestner, R. J.; Frost, J. C.; Stair, P. C.; Van Hove, M. A.;

Somorjai, G. A. Surf. Sci. 1982, 116, 85.(13) Zaera, F. Langmuir 1996, 12, 88.(14) Zaera, F. J. Catal. 1990, 121, 318.(15) Neimantsverdriet, J. W.; Borg, H. G.; van Hardeveld, R. M.; van

Saten, R. A. Recueil TraVaux Chim. Pays-Bas 1996, 115, 486.(16) Guo, X.-C.; Madix, R. J. J. Catal. 1995, 155, 336.(17) Vasquez, N., Jr.; Madix, R. J. J. Catal. 1998, 178, 234.(18) Zaera, F. J. Phys. Chem. 1990, 94, 5090.(19) Janssens, T. V. W.; Zaera, F. Surf. Sci. 1995, 344, 77.(20) Janssens, T. V. W.; Stone, D.; Hemminger, J. C.; Zaera, F. J.

Catal. 1998, 177, 284.(21) Zaera, F.; Chrysostomou, D. Surf. Sci. 2000, 457, 89.(22) Lee, I.; Zaera, F. J. Phys. Chem. B 2005, 109, 2745.(23) Lee, I.; Zaera, F. J. Phys. Chem. C 2007, 111, 10062.(24) Salmeron, M.; Somorjai, G. A. J. Phys. Chem. 1982, 86, 341.(25) Lee, I.; Zaera, F. J. Am. Chem. Soc. 2005, 127, 12174.(26) Morales, R.; Zaera, F. J. Phys. Chem. B 2006, 110, 9650.(27) Morales, R.; Zaera, F. J. Phys. Chem. C. in press.(28) Hwu, H. H.; Eng, J.; Chen, J. G. J. Am. Chem. Soc. 2002, 124,

702.(29) Khan, N. A.; Chen, J. G. J. Vac. Sci. Technol. A 2003, 21, 1302.(30) Murillo, L. E.; Khan, N. A.; Chen, J. G. Surf. Sci. 2005, 594, 27.(31) Burke, M. L.; Madix, R. J. J. Am. Chem. Soc. 1991, 113, 3675.(32) Burke, M. L.; Madix, R. J. Surf. Sci. 1990, 237, 20.(33) Zaera, F. J. Phys. Chem. 1990, 94, 8350.(34) Zaera, F. Catal. Lett. 2003, 91, 1.(35) Shaikhutdinov, Sh.; Heemeier, M.; Baumer, M.; Lear, T.; Lennon,

D.; Oldman, R. J.; Jacson, S. D.; Freund, H.-J. J. Catal. 2001, 200, 330.(36) Shaikhutdinov, Sh.K.; Frank, M.; Baumer, M.; Jackson, S. D.;

Oldman, R. J.; Hemminger, J. C.; Freund, H.-J. Catal. Lett. 2002, 80, 115.(37) Doyle, A. M.; Shaikhutdinov, Sh.K.; Jackson, S. D.; Freund, H.-

J. Angew. Chem., Int. Ed. 2003, 42, 5240.(38) Doyle, A. M.; Shaikhutdinov, Sh.K.; Freund, H.-J. J. Catal. 2004,

223, 444.(39) Doyle, A. M.; Shaikhutdinov, Sh.K.; Freund, H.-J. Angew. Chem.,

Int. Ed. 2005, 44, 629.(40) Thomson, S. J.; Webb, G.; Chem, J. Chem. Commun. 1976, 526.(41) Davis, S. M.; Zaera, F.; Somorjai, G. A. J. Catal. 1982, 77, 439.(42) Zaera, F. Chem. Rec. 2005, 5, 133.(43) Twigg, G. H. Disc. Faraday Soc. 1950, 8, 152.

(44) Zaera, F.; Somorjai, G. A. In Hydrogen Effects in Catalysis:Fundamentals and Practical Applications; Paal, Z., Menon, P. G., Eds.;Marcel Dekker: New York, 1988; pp 425-447.

(45) Campbell, C. T. AdV. Catal. 1989, 36, 1.(46) Arafa, E. A.; Webb, G. Catal. Today 1993, 17, 411.(47) Houzvicka, J.; Pestman, R.; Ponec, V. Catal. Lett. 1995, 30, 289.(48) Cremer, P.; Stanners, C.; Niemantsverdriet, J. W.; Shen, Y. R.;

Somorjai, G. Surf. Sci. 1995, 328, 111.(49) Janssens, T. V. W.; Zaera, F. J. Phys. Chem. 1996, 100, 14118.(50) Zaera, F.; French, C. R. J. Am. Chem. Soc. 1999, 121, 2236.(51) Kesmodel, L. L.; Gates, J. A. Surf. Sci. 1981, 111, L747.(52) Beebe, T. P., Jr.; Yates, J. T., Jr. J. Am. Chem. Soc. 1986, 108,

663.(53) Kesmodel, L. L.; Dubois, L. H.; Somorjai, G. A. Chem. Phys.

Lett. 1978, 56, 267.(54) Skinner, P.; Howard, M. W.; Oxton, I. A.; Kettle, S. F. A.; Powell,

D. B.; Sheppard, N.; Chem, J. Faraday Trans. II 1981, 77, 1203.(55) Davis, S. M.; Zaera, F.; Gordon, B.; Somorjai, G. A. J. Catal.

1985, 92, 240.(56) Zaera, F. J. Am. Chem. Soc. 1989, 111, 4240.(57) Ohtani, T.; Kubota, J.; Kondo, J. N.; Hirose, C.; Domen, K. J.

Phys. Chem. B 1999, 103, 4562.(58) Ko, M. K.; Frei, H. J. Phys. Chem. B 2004, 108, 1805.(59) Koel, B. E.; Bent, B. E.; Somorjai, G. A. Surf. Sci. 1984, 146,

211.(60) Nieskens, D. L. S.; van Bavel, A. P.; Curulla Ferre, D.;

Niemantsverdriet, J. W. J. Phys. Chem. B 2004, 108, 14541.(61) Greenlief, C. M.; Radloff, P. L.; Zhou, X. L.; White, J. M. Surf.

Sci. 1987, 191, 93.(62) Marinova, T. S.; Kostov, K. L. Surf. Sci. 1987, 181, 573.(63) Zaera, F.; Hall, R. B. Surf. Sci. 1987, 180, 1.(64) Stuve, E. M.; Madix, R. J.; Brundle, C. R. Surf. Sci. 1985, 152/

153, 532.(65) Liu, Z.-M.; Zhou, X.-L.; Buchanan, D. A.; Kiss, J.; White, J. M.

J. Am. Chem. Soc. 1992, 114, 2031.(66) Zaera, F.; Bernstein, N. J. Am. Chem. Soc. 1994, 116, 4881.(67) Hatzikos, G. H.; Masel, R. I. Surf. Sci. 1987, 185, 479.(68) Ormerod, R. M.; Lambert, R. M.; Hoffmann, H.; Zaera, F.; Wang,

L. P.; Bennett, D. W.; Tysoe, W. T. J. Phys. Chem. 1994, 98, 2134.(69) Demuth, J. E. Surf. Sci. 1978, 76, L603.(70) Stroscio, J. A.; Bare, S. R.; Ho, W. Surf. Sci. 1984, 148, 499.(71) Zaera, F.; Hall, R. B. J. Phys. Chem. 1987, 91, 4318.(72) Backman, A. L.; Masel, R. I. J. Phys. Chem. 1990, 94, 5300.(73) Nishijima, M.; Yoshinobu, J.; Sekitani, M.; Onchi, M. J. Chem.

Phys. 1989, 90, 5114.(74) Koestner, R. J.; Frost, J. C.; Stair, P. C.; van Hove, M. A.;

Somorjai, G. A. Surf. Sci. 1982, 116, 85.(75) Avery, N. R.; Sheppard, N. Surf. Sci. Lett. 1986, 169, L367.(76) Chesters, M. A.; De La Cruz, C.; Gardner, P.; McCash, E. M.;

Pudney, P.; Shahid, G.; Sheppard, N. J. Chem. Soc., Faraday Trans. 1990,86, 2757.

(77) Cremer, P. S.; Su, X.; Shen, Y. R.; Somorjai, G. A. J. Phys. Chem.1996, 100, 16302.

(78) Bent, B. E.; Mate, C. M.; Crowell, J. E.; Koel, B. E.; Somorjai,G. A. J. Phys. Chem. 1987, 91, 1493.

(79) Zaera, F.; Chrysostomou, D. Surf. Sci. 2000, 457, 71.(80) Pettiette-Hall, C. L.; Land, D. P.; McIver, R. T., Jr.; Hemminger,

J. C. J. Am. Chem. Soc. 1991, 113, 2755.(81) Henn, F. C.; Diaz, A. L.; Bussell, M. E.; Hugenschmidt, M. B.;

Domagala, M. E.; Campbell, C. T. J. Phys. Chem. 1992, 96, 5965.(82) Xu, C.; Koel, B. E.; Newton, M. A.; Frei, N. A.; Campbell, C. T.

J. Phys. Chem. 1995, 99, 16670.(83) Scoggins, T. B.; White, J. M. J. Phys. Chem. B 1997, 101, 7958.(84) Bratlie, K. M.; Flores, L. D.; Somorjai, G. A. Surf. Sci. 2005,

599, 93.(85) Zhdanov, V. P.; Kasemo, B. Surf. Sci. Rep. 2000, 39, 25.(86) Henry, C. R. Surf. Sci. Rep. 1998, 31, 231.(87) Baumer, M.; Freund, H.-J. Prog. Surf. Sci. 1999, 61, 127.(88) Campbell, C. T.; Grant, A. W.; Starr, D. E.; Parker, S. C.; Bondzie,

V. A. Top. Catal. 2001, 14, 43.(89) St, T. P.; Goodman, D. W. Top. Catal. 2000, 13, 5.(90) Schalow, T.; Brandt, B.; Laurin, M.; Starr, D.; K., Sh.; Shaikhut-

divov; Libuda, J.; Freund, H.-J. Catal. Lett. 2006, 107, 189.(91) Libuda, J.; Meusel, I.; Hartmann, J.; Freund, H.-J. ReV. Sci.

Instrum. 2000, 71, 4395.(92) Ofner, H.; Zaera, F. J. Phys. Chem. 1997, 101, 396.(93) Brown, W. A.; Kose, R.; King, D. A. Chem. ReV. 1998, 98, 797.(94) Weiss, W.; Ranke, W. Prog. Surf. Sci. 2002, 70, 1.(95) Lemire, C.; Meyer, R.; Henrich, V.; Shaikhutdinov, S. K.; Freund,

H.-J. Surf. Sci. 2004, 572, 103.(96) Schalow, T.; Brandt, B.; Laurin, M.; Schauermann, S.; Libuda,

J.; Freund, H.-J. J. Catal. 2006, 242, 58.


(97) Kuhrs, C.; Arita, Y.; Weiss, W.; Ranke, W.; Schlogl, R. Top. Catal.2001, 14, 111.

(98) Zscherpel, D.; Ranke, W.; Weiss, W.; Schlogl, R. J. Chem. Phys.1998, 108, 9506.

(99) Joseph, Y.; Wuhn, M.; Niklewski, A.; Ranke, W.; Weiss, W.; Woll,Ch.; Schlogl, R. Phys. Chem. Chem. Phys. 2000, 2, 5314.

(100) Manzanares, I. C.; Blunt, V. M.; Peng, J. J. Phys. Chem. 1993,97, 3994.

(101) McKean, D. C.; Mackenzie, M. W.; Morrisson, A. R.; Lavalley,J. C.; Janin, A.; Fawcett, V.; Edwards, H. G. M. Spectrochim. Acta, PartA 1985, 41A, 435.

(102) Barns, A. J.; Howells, J. D. R. J. Chem. Soc., Faraday Trans. 21973, 69, 532.

(103) Avery, N. R.; Sheppard, N. Proc. R. Soc. London A 1986, 405,27.

(104) Stuve, E. M.; Madix, R. J. J. Phys. Chem. 1985, 89, 3183.(105) de la Cruz, C.; Sheppard, N. AdV. Catal. 1996, 41, 1.(106) Hoffmann, F. M. Surf. Sci. Rep. 1983, 3, 107.(107) Zaera, F. In Encyclopedia of Chemical Physics and Physical

Chemistry; Moore, J. H., Spencer, N. D., Eds.; IOP Publishing Inc.:Philadelphia, 2001; Vol. 2, pp 1563-1581.

(108) Bertolini, J. C.; Cassuto, A.; Jugnet, Y.; Massardier, J.; Tardy,B.; Tourillon, G. Surf. Sci. 1996, 349, 88.

(109) Tourillon, G.; Cassuto, A.; Jugnet, Y.; Massardier, J.; Bertolini,J. C. J. Chem. Soc., Faraday Trans. 1996, 92, 4835.

(110) Chesters, M. A.; Horn, A. B.; Ilharco, L. M.; Ransley, A. I.;Sakakini, B. H.; Veckerman, J. C. Surf. Sci. 1991, 251/252, 291.

(111) Wolter, K.; Seiferth, O.; Kuhlenbeck, H.; Baumer, M.; Freund,H.-J. Surf. Sci. 1998, 399, 190.

(112) Frank, M.; Baumer, M. Phys. Chem. Chem. Phys. 2000, 2, 4265.(113) Hollins, P. Surf. Sci. Rep. 1992, 16, 51.(114) Schauermann, S.; Hoffmann, J.; Johanek, V.; Hartmann, J.; Libuda,

J.; Freund, H.-J. Angew. Chem., Int. Ed. 2002, 41, 2532.(115) Schauermann, S.; Hoffmann, J.; Johanek, V.; Hartmann, J.; Libuda,

J.; Freund, H.-J. Catal. Lett. 2002, 84, 209.(116) Yudanov, I. V.; Neyman, K. M.; Rosch, N. Phys. Chem. Chem.

Phys. 2006, 8, 2396.(117) Stacchiola, D.; Calaza, F.; Zheng, T.; Tysoe, W. T. J. Mol. Catal.

A 2005, 228, 35.(118) Ch.Yoon, M. X. Yang, G.A. Somorjai. J. Catal. 1998, 176, 35.(119) Bent, B. E. Chem. ReV. 1996, 96, 1361.(120) Wilde, M.; Fukutani, K.; Naschitzki, M.; Freund, H.-J. Phys. ReV.

B 2008, 77, 113412.(121) Wilde, M.; Fukutani, K.; Ludwig, W.; Brandt, B.; Fischer, J.-H.;

Schauermann, S.; Freund, H.-J. Angew. Chem., submitted.

JP800205J


Documents

Isomerization and Hydrogenation of cis-2-Butene on Pd …w0.rz-berlin.mpg.de/hjfdb/pdf/515e.pdf · · 2011-11-09Isomerization and Hydrogenation of cis-2-Butene on Pd Model Catalyst